76 Rebecca Roeder
variation across apparent time within gender. These ¿ ndings are consistent
with other studies that show sustained resistance to accommodation to local
norms by members of ethnic minority groups.
The one exception is that Mexican American women under the age of 25
who were born and raised in Lansing show the same F1/F2 values for /æ/ as
the young Detroit women who were used as a control group. These speak-
ers also show accommodation to the off-glide that is characteristic of NCS
/æ/, whereas ¿ rst generation speakers do not. It is, therefore, likely that any
effects on vowel pronunciation from adjacent consonantal context that are
found in the speech of these young women in particular represent features of
the NCS—not features from Spanish.
Of the 32 speakers analyzed in the larger study, half are both native
speakers of English and lifetime residents of Lower Michigan. This half was
chosen for further analysis of the inÀ uence of conditioning phonetic environ-
ment on the pronunciation of /æ/. Respondents for this part of the study con-
sisted of ten men, ages 14 to 71, and six women, ages 14 to 23. All but two are
second or third generation residents of Lansing. Of the remaining two, one is
a 35-year-old man who has only lived in Lansing for 10 years, but has lived
in south central Michigan his entire life. The other is a 71-year-old man who
came to Lansing at the age of 3.
2.3 Methodology
All interviews were conducted entirely in English. Analog recordings were
made using a Marantz PMD201 portable cassette recorder for some speakers, a
Marantz PMD222 for the rest, and an AT831b Audio-technica uni-directional
clip-on microphone. The recordings were then digitized to 16-bit, 10,000 Hz
digital format using the acoustic software Praat. First and second formant
measurements were taken through Praat, using the sociophonetic software
program Akustyk. When possible, the vowel was measured during the steady
state. For diphthongs, a single measurement was taken just after the percep-
tual end of the transition from the preceding consonant. The larger study used
a Nearey normalization procedure (without F3) to produce regularized data,
whereas the phonetic conditioning results discussed in the following sec-
tion are based on raw data. Analysis was performed on 31 words containing
/æ/ as the stressed vowel, which were taken from recordings of each speaker
reading a longer wordlist as part of a session that included the reading of a
short passage, a dialect perception test, and a conversational question-based
sociolinguistic interview.